Free Tool · No Login Required

HNDL Risk Scanner

Paste a HuggingFace model URL to get an instant quantum risk assessment. Free, no login required.

Results are public, shareable, and get a permanent URL + embeddable badge.

Most scanned

mistralai/Mistral-7B-Instruct-v0.2
12 scans
CRITICAL 85
Qwen/Qwen2.5-7B-Instruct
11 scans
CRITICAL 90
deepseek-ai/DeepSeek-R1
10 scans
CRITICAL 100
google/gemma-2-9b-it
10 scans
CRITICAL 85
openai/whisper-large-v3
10 scans
CRITICAL 90
black-forest-labs/FLUX.1-schnell
10 scans
CRITICAL 85
google/gemma-4-26B-A4B-it
10 scans
CRITICAL 85
microsoft/Phi-4
9 scans
CRITICAL 80

Recently scanned

microsoft/Phi-4
2026-05-10 17:34:16 UTC
CRITICAL 80
Qwen/Qwen2.5-7B-Instruct
2026-05-10 17:27:26 UTC
CRITICAL 90
deepseek-ai/DeepSeek-R1
2026-05-10 17:27:17 UTC
CRITICAL 100
google/gemma-2-9b-it
2026-05-10 17:24:19 UTC
CRITICAL 85
openai/whisper-large-v3
2026-05-10 17:22:12 UTC
CRITICAL 90
mistralai/Mistral-7B-Instruct-v0.2
2026-05-10 17:22:03 UTC
CRITICAL 85
black-forest-labs/FLUX.1-schnell
2026-05-10 17:21:54 UTC
CRITICAL 85
google/gemma-4-26B-A4B-it
2026-05-10 17:18:20 UTC
CRITICAL 85

How HNDL Risk Works

Harvest Now, Decrypt Later (HNDL) is a strategy where adversaries collect encrypted data today, anticipating future quantum computers will break current encryption. AI models are high-value targets because they represent billions in training investment, proprietary architectures, and sensitive training data.

Risk Level Score Description Action
CRITICAL 80-100 Large, high-value model with proprietary data. Immediate quantum threat. Migrate to PQC immediately
HIGH 60-79 Significant model with broad adoption. Elevated risk profile. Plan PQC migration within 6 months
MEDIUM 40-59 Moderate-sized model with standard risk factors. Monitor and prepare migration plan
LOW 0-39 Smaller model or lower target value. Standard precautions apply. Follow PQC best practices

Frequently Asked Questions

What is Harvest Now Decrypt Later?
HNDL is a strategy where adversaries intercept and store encrypted data today, with the intention of decrypting it once sufficiently powerful quantum computers become available. For AI models, this means that model weights, training data, and proprietary architectures transmitted or stored with classical encryption could be compromised in the future. The risk is proportional to how valuable the model will remain when quantum decryption becomes feasible.
When will quantum computers break current encryption?
Leading estimates from NIST and major research institutions suggest that cryptographically relevant quantum computers (CRQC) could appear between 2029 and 2035. However, the timeline is uncertain. Some experts believe it could be sooner. The important point is that data intercepted today will be vulnerable whenever that threshold is crossed, making proactive protection essential for high-value assets like AI models.
How do I protect my models?
1. Register on QuantaMrkt: Submit your model to get PQC-native signatures using ML-KEM and ML-DSA (FIPS 203/204 compliant).
2. Verify integrity: Use our Shield Registry to verify that your model files have not been tampered with.
3. Migrate encryption: Use the QuantaMrkt Migrator to transition your model storage and distribution to post-quantum cryptography.
4. Monitor: Set up continuous monitoring to detect unauthorized access or distribution of your model.